% GAUTIER LE BIHAN - 2020
% Replication files for "Shocks vs Menu Costs: Patterns of Price Rigidity in an Estimated Multi-Sector 
% Menu-Cost Model ?" Review of Economics and Statistics
%
% Estimation of product level estimate where the aggregate sector is
% 3-sector model 
% inputs for Figure O in Appendix

addpath('..\..\Utilities')  
load actual_moments_k

load param_3sectors
param0=param_3sectors;


param_3sectors=zeros(227,10);
%dist=[ param0(:,4) param0(:,8) param0(:,12)]
%[m,i] = min(dist(:,:));
 for jj=1:227;
     
    p00 =param0(jj,1);%param0(jj,1);  
     mu_c0 =param0(jj,2);%pa&ram0(jj,2)+param0(jj,2)/5 ;
     sig_eps_a0=param0(jj,3);%param0(jj,3);
     %rho_a=ones(4,1)*0.6946
weight_j=actual_moments_k(jj,12)/100;

p0_init = p00;
mu_c_init = mu_c0;
sig_eps_a_init = sig_eps_a0;
rho_a_init = 0;


% Adjustement of initial search area (beyond 5 percent)

power_p0 = 1.0;
power_mu_c = 1.;
power_sig_eps_a = 1.;
power_rho_a = 1;

% save SMM

% create vector for initial parameter values
vec0 = [p00; mu_c0; sig_eps_a0];
vec0
% Keep track of results
conv_SMM = zeros(0,3);
xopt_SMM = 100000; % start iteration
conv_opt = zeros(3,6);
SMM_count = 0; % counter

% tolerance level
tol_SMM = 5e-3;



save('SMM','vec0','p00','mu_c0','sig_eps_a0',...
    'p0_init','mu_c_init','sig_eps_a_init','rho_a_init',...
    'power_p0','power_mu_c','power_sig_eps_a','power_rho_a',...
    'conv_SMM','xopt_SMM','conv_opt','SMM_count',...
    'tol_SMM','-append', 'weight_j')
% 'power_phi','phi_init','phi0'

%% Options
%  Display, TolX, TolFun, MaxFunEvals

%% Set calibration target
% std.Y, autocorr(Y), frequency of default:
moment1 = actual_moments_k(jj,2);
moment2 = actual_moments_k(jj,3);
moment3 = actual_moments_k(jj,4);
moment4 = actual_moments_k(jj,5); 
moment5 = actual_moments_k(jj,6);



%moment6 = actual_moments(jj,7);
%moment7 = actual_moments(jj,3);


target = [moment1; moment2; moment3; moment4; moment5];
scale=[actual_moments_k(jj,7:11)'];
scale=scale.*scale*1000
save('SMM','target', 'scale','-append')
%% Start minimization routine 

%  options = optimset('MaxIter',20);%, 'TolX', 1e-4,'Tolfun', 1e-3);
%  options = optimset('TolX', 1e-4,'Tolfun', 1e-3);
options = optimset('MaxFunEvals',125);
[x,fval, exiflag] = fminsearch(@SMMmain,vec0, options);

param_3sectors(jj,:)=[x' fval exiflag target' ]

 
  save param_3sectors param_3sectors;

end